Merge AudioBench-SEA results into the leaderboard
Browse filesAdds AudioBench-SEA scores to leaderboard_data.json: 124 new datasets, 21 new task tables (ASR Burmese/Tagalog/Khmer/Lao, per-language SQA, TCQ, TLoc, Speaker Recognition, Age Prediction, Paralinguistic QA), 10 new models. SEA columns are labelled [SEA].
Only datasets that carry scores are merged; the 39 AudioBench-v1 names the SEA file declared but never evaluated are skipped, so no existing dataset moved table or was duplicated. SEA Speech Translation scores are rescaled x100 (sentence-BLEU 0-1 -> 0-100).
Bundle: sidebar nav extended with the 21 new task keys; the table row filter is relaxed from 'model has every column' to 'model has at least one column' so partially-evaluated models are not dropped (this would otherwise have removed 88 of 273 published rows); Dashboard About and Recent Updates now state that AudioBench-SEA is included.
- assets/index-DMlTR0a_.js +0 -0
- assets/index-SEA20260804.js +0 -0
- index.html +1 -1
- leaderboard_data.json +0 -0
|
The diff for this file is too large to render.
See raw diff
|
|
|
|
The diff for this file is too large to render.
See raw diff
|
|
|
|
@@ -4,7 +4,7 @@
|
|
| 4 |
<meta charset="UTF-8" />
|
| 5 |
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
|
| 6 |
<title>AudioBench Leaderboard</title>
|
| 7 |
-
<script type="module" crossorigin src="/assets/index-
|
| 8 |
<link rel="stylesheet" crossorigin href="/assets/index-DpvV3hJ3.css">
|
| 9 |
</head>
|
| 10 |
<body>
|
|
|
|
| 4 |
<meta charset="UTF-8" />
|
| 5 |
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
|
| 6 |
<title>AudioBench Leaderboard</title>
|
| 7 |
+
<script type="module" crossorigin src="/assets/index-SEA20260804.js"></script>
|
| 8 |
<link rel="stylesheet" crossorigin href="/assets/index-DpvV3hJ3.css">
|
| 9 |
</head>
|
| 10 |
<body>
|
|
The diff for this file is too large to render.
See raw diff
|
|
|